Posted in Cloud Computing, Cloud Infrastructure, Cloud Operations, Data Center Industry, Data Center Regulation, Data Center Site Selection, Michael Manos, tagged AOL, Chicago, Data Center, Data Center Regulation, Pay to Play on May 30, 2012 |
Leave a Comment »
Site Selection can be a tricky thing. You spend a ton of time upfront looking for that perfect location. The confluence of dozens of criteria, digging through fiber maps, looking at real estate, income and other state taxes. Even the best laid plans, and most thoughtful of approaches can be waylaid by changes in government, the emergence of new laws, and other regulatory changes which can put your selection at risk. I was recently made aware of yet another cautionary artifact you might want to pay attention to: Pay to Play laws and budget challenged States.
As many of my frequent readers know, I am from Chicago. In Chicago, and Illinois at large “Pay to Play” has much different connotations than the topic I am about to bring up right now. In fact the Chicago version broke out into an all out National and International Scandal. There is a great book about it if you are interested, aptly entitled, Pay to Play.
The Pay to Play that I am referring to is an emerging set of regulations and litigation techniques that require companies to pay tax bills upfront (without any kind of recourse or mediation) which then forces companies to litigate to try and recover those taxes if unfair. Increasingly I am seeing this in states where the budgets are challenged and governments are looking for additional funds and are targeting Internet based products and services. In fact, I was surprised to learn that AOL has been going through this very challenge. While I will not comment on the specifics of our case (its not specifically related to Data Centers anyway) it may highlight potential pitfalls and longer term items to take into effect when performing Data Center Site Selection. You can learn more about the AOL case here, if you are interested.
For me it highlights that lack of understanding of Internet services by federal and local governments combined with a lack of inhibition in aggressively pursuing revenue despite that lack of understanding can be dangerous and impactful to companies in this space. These can pose real dangers especially in where one site selects for their facility. These types of challenges can come into play whether you are building your own facility, selecting a colocation facility and hosting partner, or if stretched eventually where your cloud provider may have located their facility.
It does beg the question as to whether or not you have checked into the financial health of the States you may be hosting your data and services in. Have you looked at the risk that this may pose to your business? It may be something to take a look at!
Read Full Post »
Posted in AOL, AOL Data Centers, AOL MicroDataCenters, Cloud Infrastructure, Cloud Operations, Cloud Regulation, Data Center Conferences, Data Center Industry, Data Center Processes, Data Center Regulation, Michael Manos, Microsoft Data Centers, Modular Data Center, tagged BroadGroup, Christian Belady, Data Center Independence Day, Microdata Centers, Quincy Washington, Steven Wallage on May 24, 2012 |
2 Comments »
Today marked the closing lot of sessions for DataCentres2012 and my keynote session to the attendees. After sitting through a series of product, technology, and industry trend presentations over the last two days I was feeling that my conversation would at the very least be something different. Before I get to that – I wanted to share some observations from the morning.
It all began with an interesting run-down of the Data Center and infrastructure industry trends across Europe from Steve Wallage of The BroadGroup. It contained some really compelling information and highlighted some interesting divergence between the European market and the US market in terms of adoption and trends of infrastructure. It looks like they have a method for those interested to get their hand on the detailed data (for purchase) if you are interested. The parts I found particularly industry was the significant slow down of the Wholesale data center market across Europe while Colocation providers continued to do well. Additionally the percentages of change within the customer base of those providers by category was compelling and demonstrated a fundamental shift and move of content related customers across the board.
This presentation was followed by a panel of European Thought Leaders made up mostly of those same colocation providers. Given the presentation by Wallage I was expecting some interesting data-points to emerge. While there was a range of ideas and perspectives represented by the panel, I have to say it really got me worked up and not in a good way. In many ways I felt the responses from many (not all) on the panel highlighted a continued resistance to change in thinking around everything from efficiency, to technology approach. It represented the things I despise most about about our industry at large. Namely the slow adoption of change. The warm embrace of the familiar. The outright resistance to new ideas. At one point, a woman in the front row whom I believe was from Germany got up to ask a question if the panelists had any plans to move their facilities outside of the major metros. She referenced Christian Belady’s presentation around the idea of Data as Energy and remote locations like Quincy, Washington or Lulea, Sweden. She referred to the overall approach and thinking differently as quite visionary. Now the panel could have easily have referred to the fact that companies like Microsoft, Google, Facebook and the like have much greater software level control than a colo-provider could provide. Or perhaps they could have referenced that most of their customers are limited by distance to existing infrastructure deployments due to inefficiencies in commercial or custom internally deployed applications. Databases with response times architected for in-rack or in-facility levels of response times. They did at least reference that most customers tend to be server huggers and want their infrastructure close by.
Instead the initial response was quite strange in my mind. It was to go after the ideas as “innovative” and to imply that nothing was really innovative about what Microsoft had done and the fact that they built a “mega data center” in Dublin shows that there is nothing innovative really happening. Really? The adoption of 100% Air Side economization is something everyone does? The deployment of containerized compute capacity is run of the mill? The concepts about the industrialization of compute is old-hat? I had to do a mental double take and question whether they were even listening during ANY of the earlier sessions. Don’t get me wrong, I am not trying to be an apologist for the Microsoft program, in fact there are some tenets of the program I find myself not in agreement with. However – You cannot deny that they are doing VERY different things. It illustrated an interesting undercurrent I felt during the entire event (and maybe even our industry). I definitely got the sensation of a growing gap between users requirements and their forward roadmaps and desires and what manufacturers and service providers are providing. This panel, and a previous panel on modularization really highlighted these gulfs pretty demonstrably. At a minimum I definitely walked away with an interesting new perspective on some of the companies represented.
It was then time for me to give my talk. Every discussion up until this point had really focused on technology or industry trends. I was going to talk about something else. Something more important. The one thing seemingly missing from the entire event. That is – the people attending. All the technology in the world, all of the understanding of the trends in our industry are nothing unless the people in the room were willing to act. Willing to step up and take active roles in their companies to drive strategy. As I have said before – to get out of the basement and into the penthouse. The pressures on our industry and our job roles has never been more complicated. So I walked through regulations, technologies, and cloud discussions. Using the work that we did at AOL as a backdrop and example – I really tried to drive my main point. That our industry – specifically the people doing all the work – were moving to a role of managing a complex portfolio of technologies, contracts, and a continuum of solutions. Gone are the days where we can hide sheltered in our data center facilities. Our resistance to embrace change, need to evolve with us, or it will evolve around us. I walked through specific examples of how AOL has had to broaden its own perspective and approach to this widening view of our work roles at all levels. I even pre-announced something we are calling Data Center Independence Day. An aggressive adoption of modularized compute capacity that we call MicroData Centers to help solve many of the issues we are facing as a business and the rough business case as to why it makes sense for us to move to this model. I will speak more of that in the weeks to come with a greater degree of specifics, but stressed again the need for a wider perspective to manage a large portfolio of technologies and approaches to be successful in the future.
In closing – the event was fantastic. The ability this event provides to network with leaders and professionals across the industry was first rate. If I had any real constructive feedback it would be to either lengthen sessions, or reduce panel sizes to encourage more active and lively conversations. Or both!
Perhaps at the end of the day, it’s truly the best measure of a good conference if you walk away wishing that more time could be spent on the topics. As for me I am headed back Stateside and to digging into the challenges of my day job. To the wonderful host city of Nice, I say Adieu!
Read Full Post »
Posted in AOL, AOL Data Centers, Cloud Operations, Data Center Conferences, Data Center Industry, Data Center Metrics, Infrastructure, Michael Manos, Microsoft Data Centers, Modular Data Center, Open Compute Project, Personal Thoughts, tagged AOL, Chayora, Christian Belady, data center availability, facebook, Global Switch, hp, InterXion, Jack Tison, Jay Park, John Corcoran, Kfir Godrich, Len Bosack, Michael Manos, Microsoft, Microsoft Data Center, Oliver J Jones, Panduit, Paul-Francois Cattier, Prineville, Schneider, Tom Furlong, XKL Systems on May 23, 2012 |
Leave a Comment »
We wrapped our first full day of talks here at DataCentre2012 and I have to say the content was incredibly good. A couple of the key highlights that really stuck out in my mind were the talk given by Christian Belady who covered some interesting bits of the Microsoft Data Center Strategy moving forward. Of course I have a personal interest in that program having been there for Generation1 through Generation4 of the evolutions of the program. Christian covered some of the technology trends that they are incorporating into their Generation 5 facilities. It was some very interesting stuff and he went into deeper detail than I have heard so far around the concept of co-generation of power at data center locations. While I personally have some doubts about the all-in costs and immediacy of its applicability it was great to see some deep meaningful thought and differentiation out of the Microsoft program. He also went into a some interesting “future” visions which talked about data being the next energy source. While he took this concept to an entirely new level I do feel he is directionally correct. His correlations between the delivery of “data” in a utility model rang very true to me as I have long preached about the fact that we are at the dawning of the Information Utility for over 5 years.
Another fascinating talk came from Oliver J Jones of a company called Chayora. Few people and companies really understand the complexities and idiosyncrasies of doing business let alone dealing with the development and deployment of large scale infrastructure there. The presentation done by Mr. Jones was incredibly well done. Articulating the size, opportunity, and challenges of working in China through the lens of the data center market he nimbly worked in the benefits of working with a company with this kind of expertise. It was a great way to quietly sell Chayora’s value proposition and looking around the room I could tell the room was enthralled. His thoughts and data points had me thinking and running through scenarios all day long. Having been to many infrastructure conferences and seeing hundreds if not thousands of presentations, anyone who can capture that much of my mindshare for the day is a clear winner.
Tom Furlong and Jay Park of Facebook gave a great talk on OCP with a great focus on their new facility in Sweden. They also talked a bit about their other facilities in Prineville and North Carolina as well. With Furlong taking the Mechanical innovations and Park going through the electrical it was a great talk to created lots of interesting questions. An incredibly captivating portion of the talk was around calculating data center availability. In all honesty it was the first time I had ever seen this topic taken head on at a data center conference. In my experience, like PUE, Availability calculations can fall under the spell of marketing departments who truly don’t understand that there SHOULD be real math behind the calculation. There were two interesting take aways for me. The first was just how impactful this portion of the talk had on the room in general. There was an incredible amount of people taking notes as Jay Park went through the equation and way to think about it. It led me to my second revelation – There are large parts of our industry who don’t know how to do this. In private conversations after their talk some people confided that had never truly understood how to calculate this. It was an interesting wake-up call for me to ensure I covered the basics even in my own talks.
After the Facebook talk it was time for me to mount the stage for Global Thought Leadership Panel. I was joined on stage by some great industry thinkers including Christian Belady of Microsoft, Len Bosack (founder of Cisco Systems) now CEO XKL Systems, Jack Tison-CTO of Panduit, Kfir Godrich-VP and Chief Technologist at HP, John Corcoran-Executive Chairman of Global Switch, and Paul-Francois Cattier-Global VP of Data Centers at Schneider Electric. That’s a lot of people and brainpower to fit on a single stage. We really needed three times the amount of time allotted for this panel, but that is the way these things go. Perhaps one of the most interesting recurring themes from question to question was the general agreement that at the end of the day – great technology means nothing without the will do something different. There was an interesting debate on the differences between enterprise users and large scale users like Microsoft, Google, Facebook, Amazon, and AOL. I was quite chagrined and a little proud to hear AOL named in that list of luminaries (it wasn’t me who brought it up). But I was quick to point out that AOL is a bit different in that it has been around for 30 years and our challenges are EXACTLY like Enterprise data center environments. More on that tomorrow in my keynote I guess.
All in all, it was a good day – there were lots of moments of brilliance in the panel discussions throughout the day. One regret I have was on the panel regarding DCIM. They ran out of time for questions from the audience which was unfortunate. People continue to confuse DCIM as BMS version 2.0 and really miss capturing the work and soft costs, let alone the ongoing commitment to the effort once started. Additionally there is the question of once you have mountains of collected data, what do you do with that. I had a bunch of questions on this topic for the panel, including if any of the major manufacturers were thinking about building a decision engine over the data collection. To me it’s a natural outgrowth and next phase of DCIM. The one case study they discussed was InterXion. It was a great effort but I think in the end maintained the confusion around a BMS with a web interface versus true Facilities and IT integration. Another panel on Modularization got some really lively discussion on feature/functionality and differentiation, and lack of adoption. To a real degree it highlighted an interesting gulf between manufacturers (mostly represented by the panel) who need to differentiate their products and the users who require vendor interoperability of the solution space. It probably doesn’t help to have Microsoft or myself in the audience when it comes to discussions around modular capacity. On to tomorrow!
Read Full Post »
This morning I sat in on Christian Belady’s presentation at DataCentre2012. I will post small blips about things that interest me as the conference continues.
Both Christian and Laurent Verneray of Schneider each identified 5 megatrends. Interestingly while there were common themes between them at a high level, they attacked the trends from different altitudes of the data centre problem space. Both discussed the coming pressure on water as a resource.
He then went on to talk about the Microsoft Data Center strategy. Its probably worth a specific post from me on my observations on their evolution.
Read Full Post »
Posted in AOL, Cloud Cat and Mouse, Cloud Computing, Cloud Infrastructure, Cloud Operations, Data Center Conferences, Data Center Industry, Green IT, Michael Manos, tagged datacentre, DataCentres2012, France, Michael Manos, Nice on May 21, 2012 |
1 Comment »
This week I am headed off to France to be a keynote speaker at DataCentres2012. In my opinion this event is the pre-eminent infrastructure and operations conference across the whole of Europe if not the world. Regularly attracting the best speakers and infrastructure professionals in the industry (myself excluded of course – perhaps they are looking for some comic relief?) along with an incredible list of attendees which is a veritable who’s who of our industry world-wide.
By the looks of it, Cloud and Energy concerns will be the topic of many of the conversations. No doubt many will focus on the impact of technology, its usefulness, features, and capabilities. But those of you who have heard me speak before know, that I believe there is a larger more personal story – for both the professional and the companies they represent. The problems we are facing in the industry today are complicated, multi-faceted, and deep-rooted in the past of our own decisions or the decisions of our predecessors. We sometimes think technology alone is the salve for all ills. This is no more true than the purchase of a pen being the solution to writers block.
My talk will center on this multi-faceted problem space. I will use real world examples of how I have, and am tackling those issues. Who knows? I might even let loose some of the top secret work we have been doing internally to position us for the future.
As always – If you happen to be at the conference or in Nice - Don’t be a stranger!
Read Full Post »
Posted in AOL, AOL Data Centers, Cloud Computing, Cloud Operations, Data Center Industry, Data Center Metrics, Green IT, Online Operations, Personal Thoughts, Uptime Institute, tagged Aaron Lake, AOL, AOL Cloud, Christy Abramson, Julie Edwards, Lee Ann Macerelli, Rachel Paiva, Symposium 2012, Uptime Institute on May 14, 2012 |
1 Comment »
This week my teams have descended upon the Uptime Institute Symposium in Santa Clara. The moment is extremely bittersweet for me as this is the first Symposium in quite sometime I have been unable to attend. With my responsibilities expanding at AOL beginning this week there was simply too much going on for me to make the trip out. It’s a down right shame too. Why?
We (AOL) will be featured in two key parts at Symposiums this time around for some incredibly ground breaking work happening at the company. The first is a recap of the incredible work going on in the development of our own cloud platforms. Last year you may recall that we were asked to talk about some of the wins and achievements we were able to accomplish with the development of our cloud platform. The session was extremely well received. We were asked to come back, one year on, to discuss about how that work has progressed even more. Aaron Lake, the primary developer of our cloud platforms and my Infrastructure Development Team, will be talking on the continued success, features, and functionality, and the launch of our ATC Cloud Only Data Center. Its been an incredible break neck pace for Aaron and his team and they have delivered world-class capabilities for us internally.
Much of Aaron’s work has also enabled us to win the Uptime Institutes First Annual Server Round Up Award. I am especially proud of this particular honor as it is the result of an amazing amount of hard work within the organization on a problem faced by companies all over the planet. Essentially this is Operations Hygiene at a huge scale, getting rid of old servers, driving consolidation, moving platforms to our cloud environments and more. This talk will be lead by Julie Edwards, our Director of Business Operations and Christy Abramson, our Director of Service Management. Together these two teams led the effort to drive out “Operational Absurdities” and our “Power Hog” programs. We have sent along Lee Ann Macerelli and Rachel Paiva who were the primary project managers instrumental in making this initiative such a huge success. These “Cowgirls” drove an insane amount of work across the company resulting in over 5 million dollars of un-forecasted operational savings, proving that there is always room for good operational practices. They even starred in a funny internal video to celebrate their win which can be found here using the AOL Studio Now service.
If you happen to be attending Symposium this year feel free to stop by and say hello to these amazing individuals. I am incredibly proud of the work that they have driven within the company.
Read Full Post »
Posted in AOL, Cloud Infrastructure, Data Center Industry, Data Center Metrics, IPV6, tagged Akamai, AOL, AOL.com, Engadget, IPV6, IPV6 Launch Day, IPV6 Transition, Lessons Learned, Mapquest, Michael Manos on May 7, 2012 |
Leave a Comment »
Today we celebrated going live with IPV6 versions of many of our top rated sites. The work was done in advance of our participation in the IPV6 Launch Day. For the uninitiated, IPv6 Launch Day is the date where major sites will begin to have their websites publicly available in the new Internet numbering scheme and is currently set for June 6, 2012. As many of you likely know the IPv4 space which has served the Internet so well since its inception is running out of unique addresses. I am especially proud of the fact that we are the first of the largest Internet players to achieve this unique feat. In fact three of our sites occupy slots in the Top 25 Sites in the ranking including www.aol.com, www.engadget.com, and www.mapquest.com. As with all things there are some interesting caveats. For example – Google is IPv6 enabled for some ISPs, but not all. I am specifically highlighting global availability.
The journey has been an interesting one and there are many lessons learned going through these exercises. In many conversations on this topic with other large firms exploring making this move, I often hear how difficult this process appears to be and a general reluctance to even begin. Like the old saying goes even the longest journey begins with the first step.
This work was far from easy and our internal team had some great learnings in our efforts to take our sites live beginning with the World IPV6 day in 2011. Although our overall traffic levels remain pretty tiny (sustained about 4-5Mb/s) it is likely to grow as more ISPs convert their infrastructure to IPv6.
Perhaps the most significant thing I would like to share is that while migrating to the numbering system – I was very pleased to find the number of options available to companies in staging their moves to the IPv6. Companies have a host of options available to them outside of an outright full scale renumbering of their networks. There are IPv4 to IPv6 Gateways, capabilities already built into your existing routing and switching equipment that could ease the way, and even some capabilities in external service providers like Akamai that could help ease your adoption into the new space. Eventually everything will need to get migrated and you will need to have a comprehensive plan to get you there, but its nice to know that firms have a bunch of options available to assist in this technical journey.
Read Full Post »
Posted in Cloud Computing, Cloud Infrastructure, Data Center Industry, Data Center Patents, Data Center Processes, Data Center Technologies, Green IT, Infrastructure, Open Compute Project, Open Source Data Centers, Technology Leadership, Yahoo, tagged data center innovation, facebook, Glasnost, Michael Manos, OCP, open compute project, patent wars, Yahoo on May 4, 2012 |
2 Comments »
Yahoo may have just sent a cold chill across the data center industry at large and begun a stifling of data center innovation. In a May 3, 2012 article, Forbes did a quick and dirty analysis on the patent wars between Facebook and Yahoo. It’s a quick read but shines an interesting light on the potential impact something like this can have across the industry. The article, found here, highlights that :
In a new disclosure, Facebook added in the latest version of the filing that on April 23 Yahoo sent a letter to Facebook indicating that Yahoo believes it holds 16 patents that “may be relevant” to open source technology Yahoo asserts is being used in Facebook’s data centers and servers.
While these types of patent infringement cases happen all the time in the Corporate world, this one could have far greater ramifications on an industry that has only recently emerged into the light of sharing of ideas. While details remain sketchy at the time of this writing, its clear that the specific call out of data center and servers is an allusion to more than just server technology or applications running in their facilities. In fact, there is a specific call out of data centers and infrastructure.
With this revelation one has to wonder about its impact on the Open Compute Project which is being led by Facebook. It leads to some interesting questions. Has their effort to be more open in their designs and approaches to data center operations and design led them to a position of risk and exposure legally? Will this open the flood gates for design firms to become more aggressive around functionality designed into their buildings? Could companies use their patents to freeze competitors out of colocation facilities in certain markets by threatening colo providers with these types of lawsuits? Perhaps I am reaching a bit but I never underestimate litigious fervor once the proverbial blood gets in the water.
In my own estimation, there is a ton of “prior art”, to use an intellectual property term, out there to settle this down long term, but the question remains – will firms go through that lengthy process to prove it out or opt to re-enter their shells of secrecy?
After almost a decade of fighting to open up the collective industry to share technologies, designs, and techniques this is a very disheartening move. The general Glasnost that has descended over the industry has led to real and material change for the industry.
We have seen the mental shift of companies move from measuring facilities purely around “Up Time” measurements to one that is primarily more focused around efficiency as well. We have seen more willingness to share best practices and find like minded firms to share in innovation. One has to wonder, will this impact the larger “greening” of data centers in general. Without that kind of pressure – will people move back to what is comfortable?
Time will certainly tell. I was going to make a joke about the fact that until time proves out I may have to “lawyer” up just to be safe. Its not really a joke however because I’m going to bet other firms do something similar and that, my dear friends, is how the innovation will start to freeze.
Read Full Post »